| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
 |
ABSTRACT |
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
METHODS |
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Household selection. The communities were visited Mon-day through Thursday between 1 and 5 PM, when residents were less busy. Households were included if residents were present on any 1 of 3 visits. Out of 184 households, 75 households were excluded because all inhabitants worked outside the community and only returned late at night or on Sundays. The resident primarily responsible for collecting water was asked to participate and provided informed consent. Of the 109 households that met our inclusion criteria, 16 (15%) declined to participate. The protocol was approved by the ethics committee of A.B. PRISMA.
In-house stages of water storage and consumption. Households stored water for drinking, cooking, washing clothes, and other uses in containers of varying types and sizes. This first stage was defined as principal water storage. We knew from preliminary data that water was habitually taken from this container, boiled, and then stored in smaller containers for later drinking (stage 2, drinking water storage). If the household did not have boiled drinking water at the time of the visit, a subsequent visit was arranged. The final stage consisted of drinking water served in a cup (stage 3, drinking water consumption).
Sampling methods. The resident was first asked to rinse each hand for 30 seconds in a plastic (Zip-loc) bag containing 100 mL of distilled water. Five other water samples were collected in 250-mL sterile glass bottles. First, samples from each of the 3 stages described above were collected: 1) water from the principal storage container as normally collected by residents; 2) drinking water taken directly from the boiled drinking water container; 3) drinking water as served in a cup. Then the resident was asked to wash the cup as they normally would, and 2 additional samples were collected: 4) boiled drinking water, as served in the same cup after it was washed; and 5) the water used to rinse the cup after being washed for sample 4. Samples from shared community water sources were collected separately.
In samples from the principal storage stage and from shared community sources, free active chlorine was measured using the N,N-diethylphenylenediamine (DPD) colorimetric method (Hach Company, Loveland, CO), and pH was measured with Panpeha indicator strips (Sigma-Aldrich, Seelze, Germany). Water temperature was measured at each stage with a thermometer. Turbidity was not measured because the source water was free of sediment. To dechlorinate samples, 2 drops of a 3% sodium thiosulfate (Na2S2O3) solution added for each 100 mL of water. Samples were labeled, coded, placed on icepacks, and protected from light in a sealed cooler.
A cotton swab moistened with sterilized water was passed over the handle or outer surface of the vessel used to obtain cooking water from the principal storage container. A second swab was taken from the entire inner surface of an unused drinking cup. These swabs were transported to the laboratory in loosely sealed plastic tubes containing 10 mL of sterilized water.
A structured 30-minute questionnaire was administered in Spanish, collecting information on the respondent’s personal and domestic hygiene practices, water handling and usage, and sociodemographics. At the end of the interview, each household was given an explanation of the results of the chlorine test and a bottle of purified water as a gift for participation.
Laboratory methodology. All water samples were transported to the laboratory at the Peruvian University Cayetano Heredia and processed within 6 hours of sampling using membrane filtration and incubated on m-ColiBlue24 in accordance with the manufacturer’s instructions (Hach Company, Loveland, CO). Escherichia coli colonies per 100 mL of sample were enumerated based on the number of blue colonies.22 One-hundred milliliters of sample were filtered directly, except for the hand- and dish-rinsing samples, for which additional 1:2 and 1:40 dilutions were prepared and filtered.
The tubes with the swabs and solution were agitated, and then the swabs were discarded. One milliliter of the solution was then added to 9 mL of previously prepared lauryl sulfate broth, which was then incubated at 45°C for no more than 24 hours. At that time, if growth had occurred, as shown by gas bubbles and cloudiness, serial dilution was performed to obtain a 1:106 dilution, of which 0.1 mL was inoculated on EMB agar and incubated at 37°C for no more than 24 hours. After incubation, colonies with a metallic green color were counted.23,24
Statistical analysis. Multivariable logistic regression analyses were performed to examine factors associated with fecal contamination of water samples at each of the 3 stages (principal storage, drinking water storage, and drinking water consumption). Water quality was measured by a binary variable indicating whether any E. coli was detected. This measure was chosen instead of a continuous scale due to the frequency of E. coli "too numerous to count." We examined potential risk factors based on specific container characteristics: type; size; large opening; storage location; covered storage; and access of children and animals. A number of exploratory covariates measured hygiene practices related to the water vessels at each stage: time since washing of vessel; form of washing; frequency of washing; and whether the drinking cup was wet or dry at the time of sampling. Measures of water temperature, ambient temperature and humidity, and free active chlorine at the principal storage stage examined possible influences on the survival of bacteria. Boiling was assumed to remove any residual chlorine, so it was not tested at subsequent stages. The source of water, whether from a neighbor’s connection or the community standpipe, was examined at the principal storage stage. Whether the stored drinking water had been prepared as tea and the time since boiling, as a proxy variable for temperature, were included at the drinking water storage stage. Finally, the household sanitary and socioeconomic conditions were measured by the following variables: toilet type; family size; reported monthly household income; education of household head; and property ownership. All continuous variables were categorized in tertiles, and contiguous tertiles were aggregated post hoc if they had comparable frequencies of contamination. Binomial family, logistic link generalized linear models were used to calculate the odds ratios for E. coli contamination in univariate analyses and multivariable logistic regression models. Multivariable logistic regression models were built using a manual forward stepwise approach. The most significant covariates, based on the results of likelihood-ratio tests, were sequentially added to the model.
Finally, the Wilcoxon matched-pairs signed-ranks test was used to compare intrahousehold paired continuous E. coli counts between stages (i.e., E. coli counts in principal storage container versus boiled drinking water storage and drinking water storage versus consumption, within the same household). Observations from households where samples from both stages were "too numerous to count" were not included in this analysis because it was impossible to determine if the concentration of E. coli had increased, decreased, or remained the same.7 If only 1 of the matched samples was "too numerous to count," then it was given the value of the upper limit of the dilution (e.g., 200 if directly filtered). Data from all 93 houses are presented unless otherwise noted. All analyses were performed with Intercooled Stata 8.2 (StataCorp LP, College Station, TX).
|
RESULTS |
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Water collection. No study household had an in-house water connection. All water was bought from nearby households with water connections (82%) or collected from a shared standpipe (18%). Both of these outlets provided water from the same high-quality source. Two samples taken 1 month apart from the standpipe had chlorine residuals of 0.7 and 0.5 mg/L, and samples from each of the 2 houses where other participants reported buying water had residuals of 0.6 and 0.5 mg/L. All source water samples were negative for fecal contamination. All families without a connection used hoses to channel water from the source outlet directly to storage containers in their homes.
Principal water storage.
Of 93 households, 39 (42%) used large cisterns made of cement and brick or prefabricated plastic tanks, and 41 (44%) used barrels of metal or plastic to store water (Table 1
). The median reported time since filling was 3 days but ranged from 2 hours to 1 month. Water temperature ranged from 16 to 25.5°C, and, of 89 samples, all had a neutral pH of 7. Of 92 water samples, 32 (35%) had no free active chlorine residual and 44 (48%) had < 0.2 mg/L. A negative correlation was found between the chlorine residual and the days since filling (Spearman 
= –0.526, P < 0.01).
|
). Multivariable logistic regression revealed that the amount of chlorine residual and water temperature were the most significant predictors of fecal contamination (Table 2). Compared with water with no chlorine residual, water with any residual chlorine was less frequently contaminated, and higher water temperatures reduced the odds of contamination.
|
|
Drinking water storage.
All but 1 of the 93 study households reported that they normally boiled their drinking water. The 1 household that did not boil drinking water reported that this was due to the cost of kerosene. In 1 other household, boiled water was not available for testing because it had been used to prepare lemonade. Of the 91 remaining households, 39 (43%) used pitchers and 27 (30%) used kettles for storage after boiling (Table 1
). Water sample temperature ranged from 17.5 to 58°C. The median time since boiling was 7 hours, but ranged from < 1 to > 31 hours. Twenty-seven (30%) samples were contaminated with E. coli (Figure 1). In multivariable logistic regression models, boiled water not stored in a kettle was almost 8 times as likely to be contaminated, and water temperature 
30°C was associated with a lower risk of contamination (Table 2).
No difference in quality was detected between principal storage container water and stored boiled drinking water (N = 85, P = 0.16). However, in the 54 houses where the principal stored water had detectable free active chlorine, the stored boiled drinking water was significantly more contaminated (Table 3). In the houses where no free active chlorine remained in principal stored water, no significant difference was detected between this water and boiled drinking water (P = 0.78).
|
). Twelve (13%) of 93 swabs from drinking cup inner surfaces grew a geometric mean of 30 E. coli cfu/mL (range, 3–300). Domestic and personal hygiene. When asked how the cup was washed before serving water, 75 (82%) of 92 respondents reported using clean water to rinse the cup; the remainder rinsed the cup in "used" water. Eighty-nine (97%) of 92 respondents used detergent, and 80 (91%) of 88 respondents washed with a sponge instead of their hands. All respondents used "new" water for the final rinse. No one reported using hot water. Only 10 (11%) of 92 respondents used a towel to dry after washing.
A total of 64 water samples used for the final cup rinse were collected. Of these, 51 samples (80%) grew a geometric mean of 2.00 x 103 E. coli cfu/100 mL (range, 5 > 8,000). There was no significant difference in E. coli counts in the boiled water from each drinking cup before and after washing (P = 0.75).
One-minute rinses of respondents’ hands revealed extensive fecal contamination: 85 (91%) of 93 hand-rinse samples were positive with a geometric mean of 177 E. coli cfu/100 mL (range, 2 > 8,000). Hand-rinse colony counts were positively correlated with cup rinse-water colony counts in the same household (Spearman = 0.27, P = 0.03).
|
DISCUSSION |
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Our data suggest strongly that the major sources of contamination resulted from poor water storage and hygiene practices in the home. Households gathered water with low but adequate amounts of free chlorine and no fecal contamination.2 Introduction of fecal contamination during transport was minimal in this community.10,15,26 Instead of carrying water in buckets or other containers, all study households used long hoses to route water directly from a standpipe or neighbor’s spigot to household storage containers. This practice likely developed as a much easier way to collect water in the steep, rocky conditions of this community.
The absence of free active chlorine was the most significant determinant of fecal contamination at the first stage, the principal storage water container. Characteristics such as a spigot or narrow mouth reduce the rate at which chlorine volatilizes from water.27 Not surprisingly, we detected lower free chlorine levels in water stored for a longer time.10 In addition, longer storage time implies more opportunity for contamination, because hands and the handle or outer surface of collecting utensils frequently carry fecal pathogens. Furthermore, the decline in water quality between source and household has been shown to be proportionately greater when source water is clean.28 Our data suggest that a similar circumstance exists within the home. We found that, in houses where principal stored water had residual free chlorine, E. coli counts increased significantly from the principal stored water stage to that of stored boiled drinking water.
Boiling failed to ensure the quality of the water people actually drank because of frequent contamination during subsequent storage. Almost one-third of our stored boiled water samples were fecally contaminated. In addition, the cost of boiling can be prohibitive.29,30 Boiling water costs about US$0.06 per liter, and families in these communities typically boil 
5 L of water per day.31,32 Households therefore spend approximately US$110 per year, or 7% of a year’s total earnings on boiling water. Furthermore, as our results show, this investment does not ensure the safety of drinking water.
For stored boiled drinking water, container type was the strongest predictor of fecal contamination. Water is safer from contamination in containers with a small opening than in those with a wide opening.7,8,10,12,33,34 Our data support this finding, with water in kettles having the lowest rate of contamination. In addition to being sterilized in the course of boiling, kettles may be a safer alternative to pots, which contribute to scalding of children.35 Practical considerations are likely to be paramount, however. Kettles are more expensive than pots and have a smaller capacity. In addition, water must be heated for cooking and bathing as well, so a scarcity of vessels could also necessitate the transfer of boiled water to a different container. Boiling is currently promoted for household-water treatment by multilateral agencies and humanitarian groups. Our evidence indicates the need for further protocols for the safe handling and storage of boiled water or, alternatively, promotion of chemical disinfection.
Finally, our data demonstrate that much of the contamination occurs in the drinking cup. Hands are likely to be the source of the fecal contamination found in swabs from the inside of glasses and in drinking cup water. We found elevated fecal bacteria counts on participants’ hands, despite the short survival time of bacteria on skin; these bacteria could easily be transferred to dish-washing water and from there to drinking cups.36 Although no significant difference in bacterial counts was found in water samples before and after washing, the high level of contamination in water used to rinse the cups was striking. When cups were wet or moist prior to filling with water, the risk of fecal contamination was higher. Contaminated drinking glasses were implicated as a potential vehicle in an outbreak of hepatitis A, and contamination of utensils by rinse water has been shown to occur for E. coli, Salmonella, and Campylobacter.37,38 The same authors recommended higher wash- and rinse-water temperatures, but none of our respondents used hot water to wash their dishes, probably because of fuel costs.39 A more practical method to ensure clean drinking cups and other water vessels would be complete air-drying followed by storage in a covered container. Washing utensils with bleach could be an alternative, since heating water would increase costs. Hygiene improvement efforts should target dish-washing practices and other aspects of domestic water management, in addition to the standard promotion of hand washing after defecation or contact with excreta.
Our results improve our understanding of the relationship among personal hygiene, domestic hygiene, and water quality and demonstrate that fecal contamination of drinking water remains commonplace in the peri-urban communities of Lima, Peru. We offer the following recommendations based on known methods of improved water storage and point-of-use treatment in the domestic domain40: 1) if source water is microbiologically clean, then use of containers with a narrow mouth, lid, and spigot would render boiling unnecessary; 2) in communities where safe sources do not exist and boiling is already practiced, education efforts should emphasize the use of kettles and other safe storage vessels for boiled drinking water; 3) point-of-use chemical disinfection of water is cheaper, safer, and more practical than boiling and, when combined with an adequate storage vessel, is an effective means of ensuring access to safe drinking water.41 Nevertheless, the best solution remains a connection in the home providing clean, chlorinated water; all else falls short.
Received January 22, 2007. Accepted for publication June 22, 2007.
Acknowledgments: We particularly thank Brigida Rosario Jimenez Hualpa for her time and knowledge of the community, and whose assistance made this study possible. We also thank Patricia Mónica Burgos Orejuela and Milagros Zavaleta Apéstegui; Margaret Kosek for advice during study design; Paula Maguiña for administrative support; and Rina Meza and NMRCD, Peru, for the use of their equipment. We are especially grateful to the residents of Las Pampas de San Juan de Miraflores for their time and cooperation.
Financial support: This study was funded by the TG35 Training Grant, "Tutorial in Tropical Health," financed by the Fogarty International Center (grant 5T35AI007646).
* Address correspondence to William E. Oswald, A.B. PRISMA, Av. Carlos Gonzales 251, Urb. Maranga, San Miguel, Lima, Peru. E-mail: william.oswald{at}jhsph.edu 
Authors’ addresses: William E. Oswald, Lilia Cabrera, and Robert Gilman, Biomedical Research Unit, A.B. PRISMA, Av. Carlos Gonzales 251, Urb. Maranga, San Miguel, Lima 32, Peru, Telephone/Fax: +51-1-464-0221, E-mail: william.oswald{at}jhsph.edu. Andrés G. Lescano, School of Public Health and Management, Universidad Peruana Cayetano Heredia, Av. Honorio Delgado 430, Urb. Ingeniería, S.M.P., Lima, Peru, Telephone: +51-1-562-3848, ext. 144, Fax: +51-1-561-3042. Caryn Bern, Division of Parasitic Diseases, National Center for Zoonotic Vector-Borne and Enteric Diseases, Centers for Disease Control and Prevention, 4770 Buford Highway NE, Atlanta, GA 30341, Telephone: +1 770-488-7654, Fax: +1 770-488-7761. Maritza M. Calderon, Laboratorio de Investigación y Desarrollo, Universidad Peruana Cayetano Heredia, Av. Honorio Delgado 430, Urb. Ingeniería, S.M.P., Lima, Peru, Telephone: +51-1-483-2942.
Reprint requests: William E. Oswald, Biomedical Research Unit, A.B. PRISMA, Av. Carlos Gonzales 251, Urb. Maranga, San Miguel, Lima, Peru. E-mail: william.oswald{at}jhsph.edu.
|
REFERENCES |
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
